The work presented in this thesis contributes to researching honeybee monitoring applications, with the aim of providing an unobtrusive monitoring solution to extend our understanding of bee behaviour and help preserve the globally threatened honeybee population.
This study presents the design and assembly of a 5.8 GHz continuous wave (CW) radar used for unobtrusive insect monitoring from a 2 to 3 meter distance. The system was designed based on matching the radar cross section (RCS) of a 4 mm steel sphere to that of a honeybee model, which estimated the RCS of the bee using a full-wave electromagnetic (EM) simulator. A honeybee hive was monitored, resulting in readouts containing accurate Doppler shifts of flying honeybees. These were used to extract the RCS of 164 free-flying honeybees and was found to be within the range of −55 to −60 dBsm ± 3 dBsm, which was within the RCS EM simulations.
The radar was integrated on a cost-effective 45 mm × 40 mm 4-layer PCB and was later for in-phase and quadrature output, supporting the identification of positive and negative Doppler shifts. It allowed the extraction of micro-Doppler signatures and provided readouts of honeybee and bumblebee wing beats that matched their expected wing beat frequency, demonstrating the capability of operating as a radar based insect classification system.
The radar was integrated with machine learning (ML) to allow automatic classification of bee behaviour. The first ML model classified a hive’s outgoing and incoming bees with an accuracy of 87.83%. The second ML model added hovering bees as a third classification group, which achieved a classification accuracy of 93.37%, thereby demonstrating the performance of the radar.
Finally, a feasibility study of coating honeybee was performed, where the upper part of the thorax and abdomen were identified as ideal coating locations. The study estimated a maximum detection range enhancement of 4.2 metres from the original 2.4 metres through the application of 100 μm coating.
This work led to several publications and demonstrated the ability of unobtrusive insect monitoring using widely available 5.8 GHz components. It addressed a research gap in honeybee RCS at 5.8 GHz and the feasibility of increased radar range monitoring using coating that will pave the way for further insect monitoring studies.